I have a dataset with 400+ columns. The first two columns I want to exclude in this calculation but I want to include it in my final output.
For columns, 3:ncol(df), how do I normalize by group?
This is what I have right now but it gave me an error and it took very long to run at the same time:
library(BBmisc)
test<-df %>%
group_by(group) %>%
mutate_all(.vars = df[3:ncol(df)],
funs(normalize))
I want to be able to set the range from 0 to 5.
This is what my dataset looks like:
df
group week col3 col4 col5 ......
A 1 25 56 87 ......
A 2 21 34 98 ......
A 3 34 67 100 ......
B 1 11 120 1000 ......
B 2 8 340 1200 ......
B 3 2 560 2000 ......
I want to apply a normalization function from column 3 and on (not hardcoded to col3 to col5 because I have 400 columns in total) by group
We need mutate_at
df %>%
group_by(group) %>%
mutate_at(vars(-one_of("week")), normalize)
# A tibble: 6 x 5
# Groups: group [2]
# group week col3 col4 col5
# <chr> <int> <dbl> <dbl> <dbl>
#1 A 1 -0.250 0.218 -1.14
#2 A 2 -0.851 -1.09 0.429
#3 A 3 1.10 0.873 0.714
#4 B 1 0.873 -1 -0.756
#5 B 2 0.218 0 -0.378
#6 B 3 -1.09 1 1.13
As mentioned in the comments,if we are using a range of index, then should be careful with mutate_at. Currently, the indexing starts without the group_by column. So, if we wanted to pass a range of index, subtract one from the start and stop positions
df %>%
group_by(group) %>%
mutate_at(vars((3-1):(ncol(.)-1)), normalize)
# A tibble: 6 x 5
# Groups: group [2]
# group week col3 col4 col5
# <chr> <int> <dbl> <dbl> <dbl>
#1 A 1 -0.250 0.218 -1.14
#2 A 2 -0.851 -1.09 0.429
#3 A 3 1.10 0.873 0.714
#4 B 1 0.873 -1 -0.756
#5 B 2 0.218 0 -0.378
#6 B 3 -1.09 1 1.13
df <- structure(list(group = c("A", "A", "A", "B", "B", "B"), week = c(1L,
2L, 3L, 1L, 2L, 3L), col3 = c(25L, 21L, 34L, 11L, 8L, 2L), col4 = c(56L,
34L, 67L, 120L, 340L, 560L), col5 = c(87L, 98L, 100L, 1000L,
1200L, 2000L)), class = "data.frame", row.names = c(NA, -6L))